Hidden Data Transmission In MELP Coded Speech Signal Using Quantization Index Modulation

In this paper, some data hiding methods which embed secret data during the analysis of the speech signal are presented. All of the methods in question hide data while the multistage quantization of the LSF parameters is in progress. Some of the given methods have already been published in the previous studies, on the other side most of the methods are novel and introduced in this study. The methods are applied to standard speech samples. The assorted methods are compared with each other using the distortion resulted from the hidden data insertion, which are calculated in forms of weighted Euclidian distances and spectral distortions. According to the comparison results, the methods which carry out QIM at the third and fourth stages of the multistage quantizer, distort the signal as much as the transmission channel including 10^-3 bit rated white noise.